| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 1.15 MB | Adobe PDF |
Advisor(s)
Abstract(s)
Evolutionary computation is attracting attention in the energy domain as an alternative to tackle inherent mathematical complexity of some problems related to high-dimensionality, non-linearity, non-convexity, multimodality, or discontinuity of the search space. In this context, the research community launched the 2020 ”Competition on Evolutionary Computation in the Energy Domain: Smart Grid Applications” and an associated simulation framework to evaluate the performance of state-of-the-art evolutionary algorithms solving real-world problems. The competition includes two testbeds: (1) Day-ahead energy resource management problem in smart grids under uncertain environments; and (2) Bi-level optimization of end-users’ bidding strategies in local energy markets. This paper describes the general framework of the competition, the two testbeds, and the evolutionary algorithms that participated. A special section is dedicated to the winner approach, CUMDANCauchy++, a cellular Estimation Distribution Algorithm (EDA). A thorough analysis of the results reveals that, led by CUMDANCauchy++, the top three algorithms form a block of approaches all based on cellular EDAs. Moreover, for testbed 2, in which CUMDANCauchy++ did not achieve the best performance, the winner approach is also based on EDAs. The outcomes of the competition show that CUMDANCauchy++ is an effective algorithm solving both testbeds, and EDAs emerge as an algorithm class with promising performance for solving smart grid applications.
Description
Keywords
Evolutionary algorithms Estimation distribution algorithms Optimization Smart grids Statistical analysis
Citation
Publisher
Elsevier